Although clozapine has been demonstrated to be clinically superior to typical neuroleptics in refractory schizophrenia, it is also more expensive. It had been hoped that the increased costs associated with its use would be offset by decreases in the utilization of other expensive resources, especially inpatient care. All patients who had clozapine initiated during an inpatient hospitalization within the VA for schizophrenia over a 4-year period (N = 1415) were matched with a comparison group (N = 2,830) on key service utilization variables and other possible confounding demographic and clinical variables using propensity scoring-an accepted statistical method, although still relatively little used in psychiatry. By using centralized VA databases, subsequent inpatient resource utilization for the 3 years after index discharge was examined. Veterans exposed to clozapine while inpatients recorded 33 (36%) more inpatient days in the subsequent 3 years after discharge than the comparison group (124 +/- 190 days vs. 91 +/- 181 days, p = .0002). When all patients exposed to clozapine were divided according to whether they had received 1 year of clozapine treatment after discharge, those that received less than 1 year's treatment recorded significantly more inpatient days than either those maintained on clozapine or controls. These results suggest that in actual practice clozapine treatment may cost substantially more than treatment with conventional neuroleptics.